Page 260 - Designing Sociable Robots

breazeal-79017 book March 18, 2002 14:24

Grand Challenges of Building Sociable Robots 241

Foerst (1999; Foerst & Petersen, 1998) explores the questions of identity and personhood
for humanoid robots, arguing that our answers ultimately reflect our views on the nature
of being human and under what conditions we accept someone into the human commu-
nity. These questions become increasingly more poignant as humans continue to integrate
technologies into our lives and into our bodies in order to “improve” ourselves, augment-
ing and enhancing our biologically endowed capabilities. Eyeglasses and wristwatches are
examples of how widely accepted these technological improvements can become. Modern
medicine and biomedical engineering have developed robotic prosthetic limbs, artificial
hearts, cochlear implants, and many other devices that allow us to move, see, hear, and
live in ways that would otherwise not be possible. This trend will continue, with visionar-
ies predicting that technology will eventually augment our brains to enhance our intellect
(Kurzweil, 2000).
Consider a futuristic scenario where a person continues to replace his/her biological
components with technologically enhanced counterparts. Taken to the limit, is there a point
when he/she is no longer human? Is there a point where she/he is no longer a person?
Foerst urges that this is not an empirical decision, that measurable performance criteria
(such as measuring intelligence, physical capabilities, or even consciousness) should not
be considered to assign personhood to an entity. The risk of excluding some humans from
the human community is too great, and it is better to open the human community to robots
(and perhaps some animals) rather than take this risk.

13.4 Reflections and Dreams

I hope that Kismet is a precursor to the socially intelligent robots of the future. Today,
Kismet is the only autonomous robot that can engage humans in natural and intuitive
interaction that is physical, affective, and social. At times, people interact with Kismet at
a level that seems personal—sharing their thoughts, feelings and experiences with Kismet.
They ask Kismet to share the same sorts of things with them.
After a three-year investment, we are in a unique position to study how people interact
with sociable autonomous robots. The work with Kismet offers some promising results, but
many more studies need to be performed to come to a deep understanding of how people
interact with these technologies. Also, we are now in the position to study socially situated
learning following the infant-caregiver metaphor. From its inception, this form of learning
has been the motivation for building Kismet, and for building Kismet in its unique way.
In the near term, I am interested in emulating the process by which infants “learn to mean”
(Halliday, 1975). Specifically, I am interested in investigating the role social interaction
plays in how very young children (even African Grey parrots, as evidenced by the work of
Pepperberg [1990]) learn the meaning their vocalizations have for others, and how to use